Describe the bug
Set up ww3_multi on a mosaic of grids with a rotated North Pole under switch RTD. The inbound part of two-way nesting is corrupted.
Expected behavior
The nesting should behave like the regtest mww3_test_02 over a gridset outer + fine. Now change the grids from cartesian to spherical. The spherical grid is placed at the zero median and the equator and approximately overlapping the cartesian grid. See the attached input files ww3_grid_rtd_*.inp.
The initial field of spectra in the outer grid is set so that the significant wave height has a spatial pattern (Fig. 1) with a maximum of 10 m in the center, and all spectra share the same single frequency – one direction towards north-east.
By default the north pole of the grids is the standard geographic North Pole, and the behavior is as expected: After about 24 hours (Fig. 2) the pattern center has reached the inner grid and continues unaffected.
To Reproduce the error
In the namelists input (attached ww3_grid_rtd_*.inp ) set the grid north pole at e.g. the geographic equator at PLON=-90., PLAT=0.. Then the spatial pattern is figuratively ‘splashed’ on the fine grid (Fig. 3). Look also at the field of spectral peak directions (dp, Fig. 4). Pale yellow is the unaffected diagonal direction, pale blue area has direction changed 90 deg diagonally upwards, and dark brown area has direction diagonally downwards.
Screenshots
Figs. 1-4
Additional context
Fig. 1
Fig. 2
Fig. 3
Fig. 4
Attached input files ww3_grid_rtd_outer.inp, ww3_grid_rtd_fine.inp, ww3_strt_rtd.inp. Rename ww3_strt_rtd.inp to ww3_strt.inp before running. Also info_RTD about running testing with run_cmake_test.
I’d like to note that this issue can appear only under switch RTD, which is not expected to be set when e.g. using curvilinear grids.
ww3_grid_rtd_fine.inp.txt
ww3_grid_rtd_outer.inp.txt
ww3_strt_rtd.inp.txt
info_RTD.txt
Describe the bug
Set up ww3_multi on a mosaic of grids with a rotated North Pole under switch RTD. The inbound part of two-way nesting is corrupted.
Expected behavior
The nesting should behave like the regtest mww3_test_02 over a gridset outer + fine. Now change the grids from cartesian to spherical. The spherical grid is placed at the zero median and the equator and approximately overlapping the cartesian grid. See the attached input files ww3_grid_rtd_*.inp.
The initial field of spectra in the outer grid is set so that the significant wave height has a spatial pattern (Fig. 1) with a maximum of 10 m in the center, and all spectra share the same single frequency – one direction towards north-east.
By default the north pole of the grids is the standard geographic North Pole, and the behavior is as expected: After about 24 hours (Fig. 2) the pattern center has reached the inner grid and continues unaffected.
To Reproduce the error
In the namelists input (attached ww3_grid_rtd_*.inp ) set the grid north pole at e.g. the geographic equator at PLON=-90., PLAT=0.. Then the spatial pattern is figuratively ‘splashed’ on the fine grid (Fig. 3). Look also at the field of spectral peak directions (dp, Fig. 4). Pale yellow is the unaffected diagonal direction, pale blue area has direction changed 90 deg diagonally upwards, and dark brown area has direction diagonally downwards.
Screenshots
Figs. 1-4
Additional context
Fig. 1
Fig. 2
Fig. 3
Fig. 4
Attached input files ww3_grid_rtd_outer.inp, ww3_grid_rtd_fine.inp, ww3_strt_rtd.inp. Rename ww3_strt_rtd.inp to ww3_strt.inp before running. Also info_RTD about running testing with run_cmake_test.
I’d like to note that this issue can appear only under switch RTD, which is not expected to be set when e.g. using curvilinear grids.
ww3_grid_rtd_fine.inp.txt
ww3_grid_rtd_outer.inp.txt
ww3_strt_rtd.inp.txt
info_RTD.txt